4 ms·
Frankly, that's how you increase your failure rate. If a system fails, you figure out why, figure out any other failure cases that may have been missed, and you
by daeken 10y ago
Frankly, that's how you increase your failure rate. If a system fails, you figure out why, figure out any other failure cases that may have been missed, and you fix them.
You don't ditch something that largely works, and replace it with something brand new, for which you have zero data. That's how you lose rockets, cargo, and lives.
- HeyLaughingBoy 10y agoSame thing applies to software: you encounter an error, you analyze why it happened. Then ask: what other places could have this error? Can I eliminate this entire class of error by design or other method or do I need to fix them all individually? You organically end up with a more robust system that way. Don't throw out the baby with the bathwater.
- SEJeff 10y agoAnd then there is the "Did you turn it off and on again" school of thought. I've had to fight against that so hard before when the correct issue is inducing something to drop core, or fix a leaky program using valgrind.